How Do You Keep Teams Performing When Their Composition Keeps Changing?

Teams that keep changing composition do not restart their development from scratch. The advice that says they do rests on a 1965 model whose own author warned it should not be generalised to workplace teams. What a membership change actually costs is something narrower, and more manageable.

Published by MASSIVUE, an enterprise AI transformation and capability building firm. Last reviewed: August 2026.

The short answer

When people join or leave a team, the team does not drop back to stage one and rebuild everything. Most of what makes a team effective is structural, and structure stays where you put it: role definitions, performance standards, tooling, process, documented decisions and the domain knowledge base.

What leaves with people is relational. Team familiarity, meaning the accumulated experience of having worked with these specific colleagues, is the asset that degrades, along with the knowledge of who knows what, the unwritten context behind past decisions, and the coordination reflexes a group builds by repetition.

That distinction changes what you do about it. The goal is not to hold teams still, which almost no large organisation can do now anyway. The goal is to route familiarity deliberately through each change so that a reconfigured team starts from something rather than nothing.

Why enterprises are asking this now

Repeated reteaming used to be a symptom of poor planning. It is now the planned state.

Mercer's Global Talent Trends 2026, drawn from nearly 12,000 business executives, HR leaders, investors and employees, found that 98 percent of executives are planning organisational design changes over the next two years, and 65 percent expect 11 to 30 percent of their workforce to be redeployed or reskilled because of AI within that window. In the same research, C-suite confidence that their organisation is well prepared for the human-machine era fell to 51 percent in 2026 from 65 percent in 2024.

Read those findings together and the picture is specific. Nearly every large organisation intends to move people between teams over the next two years, and fewer of their leaders than before believe they are ready to do it well.

Gartner's evidence describes what that feels like from inside. In research published in July 2025, only 32 percent of mid-to-senior business leaders said the last change they led achieved healthy change adoption. An April 2025 Gartner survey of more than 2,850 employees found 79 percent reporting low trust in change. Gartner's Kayla Velnoskey described changes today as continuous, stacked on one another and highly interdependent, to the point where the nature of change has become, in her word, ungovernable.

The same research offers the counterweight. Gartner reports that organisations with better than average healthy change adoption see roughly twice the year-over-year revenue growth rate. Handling reteaming well is not a morale exercise. It tracks with financial performance.

The premise most advice rests on does not survive its own source

Search this problem and the same explanation appears almost everywhere: every time a member changes, the team goes back through forming and storming. It comes from Bruce Tuckman's 1965 paper in Psychological Bulletin, which proposed the sequence of forming, storming, norming and performing. Tuckman and Mary Ann Jensen added adjourning in 1977.

The model is genuinely useful as vocabulary. It is much weaker as a prediction about your teams, and the clearest statement of why comes from Tuckman himself.

The 1965 paper was not an experiment. It was a synthesis of a collection of 50 articles, many of them psychoanalytic studies of therapy groups and T-groups. In the paper's own discussion section, Tuckman writes that the literature cannot be considered truly representative, because therapy groups were over-represented while natural groups, the category that includes workplace teams, were under-represented. He states that the imbalance necessitates caution in generalising from this literature, and that generalisation must be limited to what the review actually covered, which was mainly sequential development in therapy groups. He adds that most of the studies observed a single group, qualitatively, through an observer who was usually the therapist or trainer.

So the foundational source for the claim that your engineering squad restarts at forming when one person joins explicitly excludes workplace teams from its own conclusions, and calls for the research in those settings that had not yet been done. Darlene Bonebright's 2010 review in Human Resource Development International traces how the model travelled from practitioner use into academic literature largely on the strength of its memorable labels.

What actually degrades when membership changes

Two bodies of evidence are more useful here because they studied working teams and measured outcomes.

The first is a study of team familiarity by Robert Huckman, Bradley Staats and David Upton, published in Management Science in 2009, using project data from an Indian software services firm. Familiarity was measured as the average number of times each member had previously worked with every other member. It had a significant positive effect on performance. Individual experience measured conventionally, such as years at the firm, did not consistently predict performance. Role-specific experience did.

That finding is more actionable than it first appears. It says the useful unit is not how senior your people are, or even how long they have been with you. It is how much history they have with each other, and how well they know the specific job they are doing.

The second is an experiment by Wendy Bedwell, published in Frontiers in Psychology in 2019, which put 165 participants into 60 teams running an emergency room simulation, with performance measured before and after a membership change. Teams that stayed intact showed significantly greater adaptive performance gains than teams that lost a member and gained a replacement. Membership change also degraded the development of shared mental models. Notably, those mental models did not statistically mediate the performance effect, so the damage is not fully explained by the team simply thinking less alike.

It is a controlled laboratory study with student participants, so the size of the effect should not be transplanted onto your organisation. The direction is the useful part, and it is consistent with the familiarity research: swapping a person out and a person in costs something real, and it costs it in the relational layer.

Team familiarity is an organisational asset, not a team property

The following is MASSIVUE's editorial synthesis of the evidence above rather than a separate research finding, and it is the reframe that makes the problem tractable.

Most organisations treat familiarity as something a team happens to accumulate if it is left alone long enough. Treated that way, it is destroyed by every reorganisation, because nobody is accountable for it and nothing records it. That is why repeated reteaming feels like starting over even when the structure around the team is untouched.

Familiarity is better understood as an asset the organisation holds, distributed across pairs of people rather than located inside a team boundary. Two colleagues who have worked together for two years still carry that history when they are assigned to different squads. The organisation owns that value. It usually just does not know where any of it sits, so it cannot avoid destroying it.

Once you hold that view, reteaming decisions change shape. Splitting a team of eight into two teams of four is not one decision, it is a choice among many possible partitions, and those partitions differ enormously in how much accumulated pair history they preserve. Today that choice is almost always made on skills coverage and headcount alone. Adding familiarity as a second criterion costs nothing and preserves an asset you have already paid for.

This is also where Heidi Helfand's work on dynamic reteaming is a better guide than the stage models. Her argument, drawn from practice in software organisations, is that team change is normal and can be done well, rather than being a failure to be minimised. The evidence above does not contradict that. It qualifies it: change is survivable, and it is cheaper when you move familiarity along with the people.

Four moves that protect performance through reteaming

1. Make familiarity an explicit input to every reteaming decision

Before a split, merge or reassignment is finalised, ask which pairs of people have substantial working history and what the proposed structure does to them. Where two options are otherwise equivalent on skills and capacity, choose the one that keeps more pairs intact. Where a break is unavoidable, know which relationship you are spending and say so.

Keeping one or two established pairs together inside a newly formed team is usually cheap, and it gives the new group a working core on day one instead of a set of strangers with a shared backlog.

2. Hold performance standards at the organisation level, not the team level

If what good looks like is renegotiated inside each team, then every reconfiguration genuinely does restart that negotiation. If definition of done, quality bars, review expectations and escalation paths are set once and apply across teams, a person moving between teams carries them across and there is nothing to rebuild.

This is the single highest-leverage structural move, because it converts a recurring cost into a one-time one.

3. Write down the context, not just the decisions

Most teams record what they decided. Few record why, what was rejected, and what constraint made the choice necessary. The why is precisely the unwritten context that leaves with a departing member, and it is what a newcomer needs in order to avoid reopening settled questions.

A short decision record attached to each significant choice, covering the options considered and the reason for the one taken, is the cheapest available insurance against the cost of churn. It also compounds, because it keeps working after the person who wrote it has moved on.

4. Design entry for the team, not only for the joiner

Standard onboarding optimises for the individual: access, tools, training, orientation. The measured cost of a membership change falls on the team, in coordination and shared understanding, so entry should be designed there too.

In practice that means naming who the joiner pairs with for the first weeks, stating explicitly what that person is accountable for so the rest of the team is not guessing, and giving the team a defined window where its delivery commitments reflect the change. Role-specific experience predicted performance in the Huckman research, so getting a joiner into a clearly defined role quickly matters more than a broad orientation programme.

What to measure

Reteaming is usually evaluated by whether the new structure looks right on the org chart, which tells you nothing about whether it worked. Four measures give a more honest reading.

None of these requires new tooling. All four can be assembled from records most organisations already keep, and they are considerably more informative than counting how many people changed seats.

Where MASSIVUE fits

MASSIVUE is an enterprise AI transformation and capability building firm. The relevance here is narrow and worth stating plainly, because the diagnosis above points at organisational design and workforce movement rather than at team-building technique.

The Mercer finding that 65 percent of executives expect 11 to 30 percent of their workforce to be redeployed or reskilled because of AI is the commercial centre of this problem. AI Workforce Transformation addresses that movement directly, covering how capability is rebuilt as people are redeployed. Enterprise Transformation addresses the structural layer above it, which is where performance standards and decision rights are set once so that individual teams stop renegotiating them.

If the question you are actually facing is who is accountable for work once AI agents are inside the team, that is a different problem with a different answer, and our guide to high-performing teams in the age of AI covers it. For the sequencing of a large change programme end to end, see our end-to-end change management playbook.

Frequently asked questions

Does a team really go back to forming and storming when a member changes?

There is no strong evidence that it does, and the source usually cited for the claim does not support it. Tuckman's 1965 paper synthesised 50 studies dominated by therapy groups, and its discussion section states that the under-representation of natural groups, the category that includes workplace teams, requires caution in generalising, limiting conclusions mainly to therapy groups. Controlled research on working teams shows something narrower: membership change reduces adaptive performance and degrades the development of shared understanding. Structural elements such as role definitions, standards and documented decisions were not what those studies measured, and they persist through a reconfiguration for the simple reason that they are held outside the team. A team that changes composition loses relational capital rather than its entire development.

Is it better to keep teams stable or to reteam deliberately?

For most large organisations the choice is no longer available. Mercer's Global Talent Trends 2026 found 98 percent of executives planning organisational design changes within two years. The evidence supports a middle position rather than either extreme: team familiarity measurably improves performance, so churn is not free, but stability is not the objective either. The objective is to make each change preserve as much accumulated working history as the structure allows, and to hold standards and context outside the team so they survive the move.

How long does a team take to recover after a membership change?

No credible research gives a universal figure, and any specific number quoted without a source should be treated with suspicion. Recovery time depends on how much pair history survived, how clearly the joiner's role is defined, and whether the team's context is written down or held in people's heads. The useful approach is to measure your own time to first meaningful delivery across successive reteams and treat that trend as the benchmark, rather than importing someone else's number.

What is team familiarity and how do you measure it?

Team familiarity is the accumulated experience members have of working with each other specifically, as distinct from their individual tenure or seniority. In the Huckman, Staats and Upton study it was operationalised as the average number of times each member had previously worked with every other member, and measured that way it significantly predicted performance while years at the firm did not. Most organisations can approximate it from project or assignment records they already hold, which is enough to compare two proposed team structures before choosing one.

How do you bring someone into an existing team without slowing the team down?

Treat it as a team event rather than an individual one. Name the colleague the joiner works alongside for the first weeks, define the specific role they are accountable for rather than orienting them broadly, and adjust the team's delivery commitments for a defined window so the cost is planned instead of absorbed silently. Role-specific experience was a consistent predictor of performance in the familiarity research, which is the argument for getting someone into a clear role quickly rather than rotating them through a general introduction.

Related MASSIVUE resources

• High-Performing Teams in the Age of AI: Who Owns the Work When Agents Do It?
• End-to-End Change Management in Singapore: A 2026 Playbook
• AI Workforce Transformation
• Enterprise Transformation
• AI Change Management: Upskilling & Reskilling

Sources

• Tuckman, B. W. (1965). Developmental sequence in small groups. Psychological Bulletin, 63(6), 384 to 399.
• Tuckman, B. W., and Jensen, M. A. C. (1977). Stages of small-group development revisited. Group and Organization Studies, 2(4), 419 to 427.
• Bonebright, D. A. (2010). 40 years of storming: a historical review of Tuckman's model of small group development. Human Resource Development International, 13(1), 111 to 120.
• Huckman, R. S., Staats, B. R., and Upton, D. M. (2009). Team familiarity, role experience, and performance: evidence from Indian software services. Management Science, 55(1). doi:10.1287/mnsc.1080.0921
• Bedwell, W. L. (2019). Adaptive team performance: the influence of membership fluidity on shared team cognition. Frontiers in Psychology, 10, 2266. doi:10.3389/fpsyg.2019.02266
• Mercer. Global Talent Trends 2026, February 2026.
• Gartner. Just 32% of business leaders report achieving healthy change adoption by employees, July 2025.
• Helfand, H. Dynamic Reteaming: The Art and Wisdom of Changing Teams (2nd ed.), O'Reilly Media.